Lufthansa Industry Solutions (LHIND), a pioneer integrator in the private 5G space, has been running a test network on a Raspberry Pi in its work labs and meeting rooms – if only just to show enterprise customers (and network vendors, if pushed) how easy it can be (with the right guidance, of course.) It is anecdotal, but also instructive, because Claudius Noack (pictured above), leading the firm’s 5G play, has seen it all – since stewarding a pair of private 5G systems into existence for Lufthansa Technik, the airline services division of Lufthansa Group, back in 2020.
It was the start of the private 5G craze, effectively, and these twin projects, to connect engine inspections (with Nokia) and cabin assemblies (with Vodafone, originally) on two sides of a giant aircraft hangar in Hamburg, were among the first to light-up the new 3.7-3.8 GHz enterprise band in Germany. Five years later, they remain touchstone deployments for the whole market. LHIND has gone on to deploy 10 private 5G networks, in total – in Europe, mostly where private spectrum is available, mostly close to other big Lufthansa hubs in countries like Switzerland and Belgium.
All of them are at least partially managed by LHIND. “It is enough revenue to make sense for us to continue down this road,” remarks Noack. The business has been running a proof-of-concept with a “big warehouse facility” close to the airport in Los Angeles, in the US, since the start of last year. It is its first client outside of Europe. “We wouldn’t have thought the public reception was so bad, but there was no signal indoors at all. We brought in two radios, and solved it all – with all the metal inside the facility. The client is just happy it has connectivity everywhere it wants it.”
The US project will go to a “fully-fledged network” this year, he says. LHIND, with 2,500 staff (“mostly programmers”), handles “digitalisation” for the parent group, and for any outside clients that want to engage. Noack’s “connectivity team” is drafted in, variously, on bigger projects. “If a client has an issue with connectivity for a new AI system, say, they talk to us,” he says. “The market is basically everything with a big-sized shop floor – all warehouses, say. We have port clients, manufacturing clients. If you’re bigger than 3,000 square metres, then private 5G works.”
All of which, a long way round, tells us, also, that if you want a proper review about progress in the private 5G market over the last five years, then Noack is a good person to ask. And he has lots of interesting stuff to say, including this Raspberry Pi anecdote. “I see it from both sides – from the client’s side, about what they need, and the supplier’s side, about what they offer,” he says. “And there’s a huge difference, compared to 2020/21. Back then, no matter if it was Nokia or Ericsson, or even the [smaller] vendors, they had products that just didn’t fit [enterprise] needs.”
He explains: “They were still [shrinking-down telco products]. Yes, they were smaller versions, but they still came with fully-fledged cores, way too-much for most clients – which are looking, typically, to connect 100 end-devices. No one is connecting a million. Our biggest project is for 1,200 devices on one campus network. Which is not what they were serving in 2021 with these big and expensive cores. Most projects are less about speed, and more about reliable and secure connectivity – for which you can probably take an open-source core from a smaller vendor.”
Hence the Raspberry Pi setup, engineered by one of Noack’s students. “He ran the core on a Raspberry Pi 5 with a cheap open-source software-defined radio – and it works to connect a phone, tablet, camera, and so on,” he says. But it is not a commercial exercise, he notes; nor is it geared to undermine the vendor community – which has multiplied in the intervening period, especially in terms of core-network provision. “The point is to show clients how easy it is – that it’s not that complex anymore; that there are different solutions, and we can help find the right ones.”
Variety and viability
But there are lessons in it, too. “When I started in 2019, I was clear that we didn’t want to be an operator. ‘We can’t be, and we will never be.’ But five years later, I can say, very easily, that, yes, we can be a mobile operator for industrial clients. And we are already. That Raspberry Pi set-up is just to put on the table, to turn off two minutes later – to show the variety in the market. You still have the big systems from the big brands, but there are so many other core suppliers in the market now. A new one writes to me every month. And you look them up, and they’re like 18 months old.”
He adds: “Which also shows you how easy it is.” Does their youth raise concerns about their longevity? “That’s the discussion with clients. They are cheaper because they’re selling a software license for the core, to go on any device with enough CPU power – whereas the big vendors supply dedicated hardware. The question is always whether they will be around in two years. Nokia and Ericsson will be, because it’s their business.” They would prefer sales to go faster, clearly, he says, but they are also “safe bets”. Plus, their “quality and KPIs” are generally higher.
But equally, the art of good consultancy is to weigh the options, the message goes. “It’s apples and oranges,” says Noack. The ecosystem knows by now how varied the customer market is – between industries, geographies, enterprises, even enterprise operations – and that half the challenge is to scope the business problem to select the business solution. As such, cheaper and simpler core solutions, even if they are only amped-up Raspberry Pi projects, have a proper place. And size alone does not always guarantee much, actually – in such a difficult customer market.
The lesson of Metaswitch should tell us that, he suggests. US core network provider Metaswitch was acquired by Microsoft in 2020 (in a double-deal with Affirmed Networks) as hype about private 5G grew feverish, only to be shuttered last summer (2024) as reality hit – that solutions and sales are not simple. “It tells you how volatile this market is – when a big player jumps in because it thinks it can do a core network [and then quits a couple of years later].” Of course, Microsoft is not a traditional telecoms vendor (even if Metaswitch was), notes Noack, as well.
So, does the Microsoft example show how complicated Industry 4.0 really is? Does it show that the hype around new enterprise tech – around 5G, plus IoT and AI, historically and provisionally; and RedCap, recently – should not be trusted? Is private 5G good business and long-term business, often, but not easy-money or an instant hit? Does it prove, as well as anything, that (eye-roll and groan): 5G ain’t gonna change the world? “I said those things in 2019 and 2020, and 2021 and 2022. 5G was never going to change the world; it is an extension to Wi-Fi, and other technologies.”
Meanwhile, over five years, private 5G systems have come down in size. Variously, they are now properly-specified, for most jobs – whether for micro campus-wide or macro region-wide deployments, or most enterprise projects in-between. “Everyone has a small system now,” says Noack. “But there is no ‘one-size-fits-all’. Maybe you need a small system, and maybe you don’t. Which is where we come into play – to find the perfect solution for your business.” Increasingly, of late, its role has also been to negotiate sharp sales practice by vendors looking to stack solutions.
He explains: “Cost is still a challenge. We help a lot of clients with the tender process, and the first quote is always too high. And then, when the vendors see they have competition, they suddenly get it down by 40 or 50 percent. Which turns customers off… Plus, they want to sell other products as well; whereas most clients just want the connectivity. They have their own IoT and AI systems – probably in some remote place where they want to connect their cameras, say. But suppliers offer the whole bundle. The products have come down in size but they are pushing all this other stuff.”
He adds: “They still think that they know the clients better than the clients know themselves. Which is never a good position to take, of course.
Inputs and outputs
What about ROI? Is it easier to telegraph, and quicker to realise? Noack responds: “It is still complicated to get a quick ROI, and rare with a single use case. Which is just the same as in 2021. You need multiple cases. It’s like a hundred thousand bucks for the first network, plus licensing and project management. So it’s a couple of hundred thousand, all-in. So you have to save $200,000 at least – in the first five years, let’s say.”
He comments: “Which is roughly what we find right now in terms of the ROI. It’s never less than a year. We have never seen that.” But he wants, also, to qualify his maths; he adds: “You really have to consider the use case and the project to get accurate figures. For a small shop floor, the numbers could be way lower.” Which goes for the size investment, and presumably for either the value or the schedule of the return – depending on the use case, again. The classic case study for LHIND is the original virtual table-inspection (VTI) project for Lufthansa Technik in Hamburg.
That setup, a single use case, served with Nokia gear, was formulated during the Covid-19 no-travel lockdown. Since formalised and expanded to at least one other European engine workshop, it enabled Lufthansa Technik to collaborate remotely with clients about engine maintenance over a 5G-connected AR video feed. Noack explains: “That was different because we could not have done these table inspections at the site in this manner – which we are still doing by the way, along with other use cases – without 5G. So it was a no-brainer.”
He goes on: “But most of the time it’s a five-year ROI, roughly. And, of course, if you tell clients that, then sometimes they look the other way. Because five years is a long time.” So that’s the reality, then – for a smart forerunner in the private 5G game. But are these stackable use cases clearer now, at least? Are they multiplying? “It is mostly about connectivity, still – and the time you save to connect again (between cells). A big use case is scanning packages and parcels, and other items, for quality assurance, going in/out of warehouses and factories.”
He explains: “It works quite well (with RFID / Wi-Fi) in small environments. But if your assets are moving around a lot, from outside to inside and vice versa, then there are always issues with Wi-Fi between access points. Even with Wi-Fi 6, which we’ve tried out. And if you have high latency, of 100 milliseconds or so, you have to log back in, and the downtime for workers and processes adds up. That is the biggest use case in most of our projects. The second one is cameras and AI detection.”
And automated mobile robots and automated guided vehicles (AMRs and AGVs), too, presumably – if we are to believe the tales the industry tells? “No, not yet. Although it’s the same story, about the handover. But I’ve seen three or four projects like this, so not many. And they have a maximum of 20 AGVs, as well. So it’s not what you see on PowerPoints or fancy videos.” Indeed in the scheme of IoT ‘things’, truth is plainer than fiction; the reality is more prosaic than the hype. And Noack tells it like it is: hard graft and real gains – if the proper care, through consultation and collaboration, is taken.
But weirdly, despite all the frank talk and myth-busting, he is holding out for reduced capability 5G (5G RedCap) in private industrial cellular networks – to explode these limited estates of 100 devices into massive estates of thousands. Responding to the prompt, he says: “I am waiting for RedCap, actually. Right now, especially for positioning, I can track anything with a battery. If it doesn’t have a real battery, then the sensor-tag will last about a week. So I can’t wait for RedCap because then we’ll have 5G-IoT networks with maybe 10,000 temperature and gas sensors, say.”
He adds: “We have so many projects where the client just wants small sensors to know where things are. The AI positioning system we’ve tried uses only three radios, but only detects about three devices. While all the others require more than three radios. And the advantage with private 5G, compared to Wi-Fi, is you have a quarter of the access points – which makes it maybe a little cheaper. But more than three radios, and the advantage is lost. You need it to be able to cover a big shop floor with only a few radios.
“So in terms of 5G-IoT positioning, the choice is between lots of radios, which is expensive, or an AI system that is not properly developed yet. We need RedCap devices [to make the latter viable] – which haven’t come across my table yet.”
